Add Your Heading Text Here
Add Your Heading Text Here
There are two forms of life 1. Nature and 2. Art . Using AI to generate Ghibli studio images is stealing both art and the nature. Alright let’s put art aside for a while, we can’t stop the theft anyway; What is the cost we are paying to laugh and say “Aww this is so cute”? How much water is being wasted and how many resources from our Mother Earth are we stealing? Just to have some fun. ChatGPT or any other AI model does not just generate images from a computer, there are large data centres in play to go deep into every single prompt and crawl through the entire World Wide Web to give us what we want. The numbers would make you think again, here it goes
Training a beast like DALL-E 2 , which pumps out images, burns through about 1,150 MWh of electricity—enough to run 400 average Indian households for a year! That spits out roughly 460 metric tons of CO2, matching the annual emissions of 100 petrol-powered autos running through Delhi. On a smaller scale, generating one Ghibli-style AI image takes 0.011 kWh—about what it takes to brew 5 cups of chai on an electric kettle. In India’s coal-heavy grid, that’s 9 grams of CO2 per image. Doesn’t sound like much, but a million images? That’s 9 metric tons—think 2 cars’ worth of emissions yearly.
Thirsty Servers:
Cooling those AI-crunching servers is a water hog. Microsoft’s 2023 data shows their global ops used 1.7 billion gallons of water, with AI workloads like image generation playing a big role. Break it down: one AI image in an Indian data centre might sip 250 ml of water for cooling—half a bottle of Bisleri! If 10,000 folks whip up images daily, that’s 2.5 lakh litres gone, enough to fill 50 tanker trucks. In a dry spot like Chennai, that’s a real pinch. New air-liquid hybrid cooling could cut that by 40%.
Data Center Crunch:
All that image-making heats up servers fast. A single data centre running at full tilt—like one in Noida—can use 20 MW daily, and AI tasks bump that up 15-20%. That’s 3-4 MW extra, or enough to power 1,200 rural homes for a day. With heat comes more cooling strain—think 1.5 million litres of water monthly just to keep things chill. In India’s IT hubs, that’s sparking debates over resources. Zenith’s energy efficiency know-how could ease that load—maybe push for solar tie-ins or smarter server schedules. OpenAI does not disclose any usage of renewable energy or if it is going in the path of Net Zero. If none of this is happening, we are unknowingly depleting a lot of our natural resources, because, OpenAI is not the only one with Large Language Models out there, every application that needs data uses energy. Google, Netflix and even you, right at this moment.
What can we do about this?
Well, one person holding back on generating images or using AI cannot change the ever evolving world where AI is the new phase of technology. Instead, we can cut down on our regular emissions by
- Trying to go Solar - be it our homes or our apartment complexes - Zenith help you with that
- Carpooling or Public Transport - ditch the solo drives when you can, it’s a simple way to trim your carbon footprint
- Cutting Down on Waste - recycle, compost, reuse—small habits that add up, and Zenith’s got tips to make it easy
- Opting for Energy-Efficient Appliances - swap out old fans and ACs for ones with better star ratings; Zenith can guide you on what works best
- Pushing for Green Workplaces - nudge your office to adopt energy-saving policies; Zenith’s expertise can back you up with solid plans
